Sierra Leone
Chapter 18
Sierra Leone’s Economic Growth Performance, 1961-2000
Chapter 19 of volume 2
Victor A. B. Davies[1]
Sierra LeoneChapter 19
Contents
1. Introduction
2. The Context
2.1 History
2.2 Periodization
3. Cross-country econometric perspective
3.1 Growth Accounting
3.2 Growth Prediction
4. Markets
4.1 The Diamonds Sector
4.2 The Financial Sector
4.3 The Labour Market
5. Key Agents
5.1 Rural households
5.2 Manufacturing Firms
6. Political Economy Analysis
6.1 Growth Collapse
6.2 Civil War, 1991-2001
7. Sierra Leone and Botswana: A Comparative Analysis
8. Conclusions
References
List of Tables
Table 1: Macroeconomic Indicators 1961-2000
Table 2: Growth Accounting
Table 3: Variables in the Regression Models
Table 4a: Estimation Results: the Hoeffler Augmented Solow Model
Table 4b: Fits and Residuals from the Hoeffler Augmented Solow Model
Table 5a: Estimation Results: The O’Connell-Ndulu Model
Table 5b: Fits and Residuals from the O’Connell-Ndulu Model
Table 6: Diamond Production and Exports
Table 7: Financial Indicators
Table 8: Sierra Leone Produce Marketing Board: Prices and purchases
Table 9: Produce Prices in Liberia and Sierra Leone
List of Figures
Figure 1: Per capita GDP (1995 US$)
Figure 2: Fiscal Revenues % GDP
Sierra LeoneChapter 19 page 1
1. Introduction
Sierra Leone is a striking case of growth failure in Africa. At independence Sierra Leone’s substantial growth opportunities included a rich endowment of diamonds and other minerals, a long coastline, a developed educational system boasting the first university in Sub-Saharan Africa, and a seemingly stable political system. However, Sierra Leone ended the 20th century a growth tragedy. Per capita GDP growth for 1961-2000 averaged -1.2% per annum, compared with about 1% for Sub-Saharan Africa. Per capita GDP in constant 1995 dollars nearly halved between 1961 and 2000, from US$260 to US$150 (Figure 1). Worse still, civil war raged from 1991 to 2001.
***Insert figure 1 near hear***
What explains Sierra Leone’s growth tragedy? The literature lacks a comprehensive attempt to answer this question. Much of the focus has been on the civil war: Abdullah (1997), Abdullah and Muana (1998), Richards (1996), Smillie et al (2000) and Davies (2000). Saylor (1967) analyses the economy in the 1960s; Clapham (1976) and Reno (1995, 1998) the political economy; and van der Laan (1965) and Zack-Williams (1995) diamond mining. The chapter fills the gap in the literature with a comprehensive analysis of Sierra Leone’s post-independence growth performance from 1961 to 2000.
The analysis comprises the following blocks: cross-country econometric analysis, markets, private agents and political economy. To set the stage we provide a historical context in Section 2, highlighting the role of divisive, growth-inhibiting ethno-regional rivalries predating independence. Furthermore, we delineate Sierra Leone’s growth history into three episodes. Macroeconomic stability characterised 1961-72, the first episode. At the start of the second episode, 1973-90, the oil shocks prompted extensive market intervention, aggravating macroeconomic imbalances. After consolidating power, the ruling All People’s Congress transformed into a kleptocracy, precipitating state collapse by the late 1980s. Civil war defined the finalepisode, 1991-2000.
Section 3 contains the cross-country econometric analysis divided into a growth accounting and prediction analyses. In the growth accounting only education per worker contributed positively to the negative growth of real GDP per worker in 1960-2000. Capital per worker and total factor productivity contributed negatively. In the growth-prediction exercise the Hoeffler (2000) and O’Connell and Ndulu (2000) models under-predict Sierra Leone’s growth. A possible reason is underestimation of investment in the Hoeffler model. Second, the Barro-Lee ratio of unproductive government expenditure, the main driver of growth under-prediction in O’Connell and Ndulu (2000), may be conceptually inappropriate for a country like Sierra Leone with huge differences in domestic and international prices.
Section 4, focusing on key factor and product markets, argues that policy distortions induced market dysfunction while diamonds have been mixed blessings, fuelling the civil war and exerting a corrupting and destabilising political and economic influence. Section 5 focuses on two key agents – rural households with about 70-90% of the national population, and manufacturing firms. An inauspicious environment facing rural households encourages subsistence and informal economic activity. Manufacturing firms face severe infrastructural constraints, resulting in a miniscule manufacturing sector accounting for only about 5% of GDP and virtually no exports.
Section 6, the political economy block, identifies the following key culprits for growth collapse and the 1991-2001 civilwar: diamonds, kleptocracy, ethno-regional rivalries and external actors. Ethno-regional rivalries aided a kleptocracy that sowed the seeds of self destruction and state collapse by informalizing and controlling access to diamonds and other markets and underminingany institutions that posed a threat. The ensuing collapse of fiscal revenues, infrastructure and government services induced growth collapse. This produced suitable conditions for rebellion – extreme poverty, widespread youth unemployment and disillusionment, and government dysfunction. Libya trained some of the first rebels while the civil war was triggered from neighbouring Liberia. Diamonds sustained the war.Section 7 compares diamond resource management, and the political economy of growth in Sierra Leone and Botswana. Section 8 concludes.
2. The Context
2.1 History
The period leading to independence on 27 April 1961 featured a major divide between the Creoles, with 2% of the population, and “indigenous,” ethnic groups. The Creoles are descendants of former African slavesfromEurope and North America,or recaptured in Africa, who were resettledby Britain in the Freetown peninsular area around 1800. In 1808 Freetownbecame the British crown colony of Sierra Leone; in 1896, the hinterlandbecame a British Protectorate. The British ruled the Protectorate indirectly, through traditional rulers, and the colony directly. This divide limited interactions between the two regions. Incipient rivalries intensified in the run-up to independence, with the Protectorate-based Sierra Leone People’s Party (SLPP) emerging dominant. After independence, the Protectorate’s electoral supremacy would bridge the Creole-Protectorate rivalry with the Creoles realising that they must join a Protectorate-led party to exert any political influence (Clapham 1976).
A new divide emerged. In 1960 Siaka Stevens formed the All People’s Congress (APC) which “soon attracted a large following, particularly from the north” (Alie 1990). After independence, the APC became the main opposition party to the ruling SLPP. Incipient rivalries deepened between the pro-APC northern regions dominated by the Temnes, and the pro-SLPP south-eastern regions dominated by the Mendes. The Temnes and Mendes each account for 30% of the population, and together with their affiliates they split the national population roughly in half, a demographic characteristic fostering ethno-regional polarity. The two political parties, SLPP and APC, would rule between them throughout the post-independence period, except during periods of military rule.
2.2 Periodization
Three episodes in the domestic environment for growth are discernible in Sierra Leone’s 1961-2000 growth history: 1961-72, 1973-90, and 1991-2000. Although growth outcomes differ markedly by episode, the criterion for identifying episodes is the environment for growth, not outcomes. In 1961-72, the macroeconomic environment was relatively stable, with single digit inflation (Table 1) and the leone, the national currency, pegged to the pound sterling. Three military coups occurred between March 1967 and the restoration of democracy in 1968. However, the coups did not fundamentally change the environment for growth.
The next episode is 1973-90. The oil price shocks precipitated a tightening of price controls and larger subsidies for imported rice and fuel, aggravating macroeconomic distortions and fostering black markets. Meanwhile the All People’s Congress, which ruled from 1968 to 1992, consolidated power and transformed the government into a kleptocracy. Government services and infrastructure gradually collapsed in the 1980s. Per-capita income went into long-term decline around the mid 1980s. Civil war consummated the economic and political implosion in 1991-2000, the last episode.
***Insert Table 1 near here***
3. Cross-country econometric perspective
This section ties in Sierra Leone’s economic growth performance with the general empirical cross-country growth literature through a growth accounting and prediction analyses.
3.1 Growth Accounting
Growth accounting decomposes growth among its proximate sources, indicating the contribution of productivity gains, and physical and human capital accumulation. Our growth accounting results are based on Collins and Bosworth (1996) using the following production function[2]:
ln y = ln A + 0.35 ln k + 0.65 ln h
where y is real GDP per worker,k is physical capital per worker, h is a labour quality index; and 0.35 and 0.65 are the assumed capital and labour elasticities. A is total factor productivity capturing technological progress but it could also capture other factors such as capacity utilization. Data on growth in physical capital per workerwas generated by applying the perpetual inventory method to national accounts investment data, using initial 1950 capital stocks, with a depreciation rate of 0.05. The labour quality index imputes a return of 7% to an additional year of schooling that the adult population attains. Educational attainment data are from Barro and Lee (2000) and Cohen and Soto (2001). Typically both series are available, and then the average is used; otherwise, the available series is used.
Table 2 shows a negative growth in real GDP per worker in Sierra Leone of -0.66% for the overall 1960-2000 period. Only education per worker generated positive growth, 0.24 percentage points, about the same as Sub-Saharan Africa’s 0.25 percentage points and slightly higher than in Latin America and the Caribbean, 0.33. By default, the contribution of education per worker tends to be low: one additional year of average schooling increases human capital by only 7%. The contribution of physical capital per worker and total factor productivity over 1960-2000 was -0.17 and -0.73 percentage points respectively. In 1960-74 productivity was the main source of growth, accounting for well over half of the (positive) growth in output per worker. Thus,investment has contributed negatively to the growth of real GDP per worker in Sierra Leone, while education per worker’s contribution has been positive but low. However, negative total factor productivity during the 1991-2000 civil war period has been the main source of negative growth in real GDP per worker.
***Insert Table 2 near here***
3.2 Growth Prediction
The drawback of growth accounting is that it cannot identify the fundamental causes of growth. Growth regression, to which we now turn, is more informative in this regard. To determine the extent to which growth regression norms explain Sierra Leone’s growth performance, we predict Sierra Leone’s growth using empirical estimates from two standard cross-country growth models: Hoeffler’s (2000) System-GMM estimates of the Augmented Solow model of Mankiw, Roemer and Weil (1992), and O’Connell and Ndulu’s (2000) reduced-form ordinary least squares model. The authors provided the empirical estimates and underlying data[3].
The two models were estimated using half-decadal data for a panel of 85 countries, of which 43 are developing and 23 are in SSA. The estimation period is 1960-1990 for the Hoeffler model and 1960-97 for the other model. The regressors in the Hoeffler model are (i) initial income; (ii) the national saving rate as proxied by the ratio of investment to GDP, both at constant international prices; (iii) replacement investment, given by 0.05 plus the population growth rate; (iv) a measure of saving for human capital accumulation,using average years of schooling achieved by the population aged 15 or more; and (v) time dummies. The dependent variable is growth in real GDP per capita at 1985 international prices.
O’Connell and Ndulu (2000) follow Barro (1991) to assess the relative contributions to growth of initial conditions, structural features, institutional characteristics and the policy environment. The dependent variable is the average growth rate of real GDP per capita, national accounts data. Real GDP per capita in international dollars and life expectancy at birth, both measured at the beginning of the relevant half-decade, capture initial conditions. Demography is captured by the age dependency ratio – the ratio of the non-working age (outside 15-64 age range) to the working age (15-64) population; and growth in potential labour force participation – the average difference between the growth rate of the population 15-65 and growth rate of total population. Terms of trade shocks are captured by the initial share of exports to GDP, multiplied by the average percentage difference between the terms of trade in each year of the half decade and the terms of trade in the initial year of the half-decade. The average growth rate of real GDP per capita among trading partners, weighted by shares in total trade, captures trading partner growth. A dummy variable taking a value of one for landlocked countries and zero otherwise, captures landlockedness. The inflation rate, the black market premium, and the Barro-Lee unproductive government spending to GDP ratio (government spending net of expenditure on education and defence), capture policy. The “Barro-Lee” ratio is the ratio of real government consumption to real GDP, both at 1985 international prices, minus the ratio of the sum of nominal government spending on defense and noncapital education expenditures to nominal GDP. Due to data constraints, Barro and Lee use nominal, rather than real, values for the ratio of spending on education and defence to GDP. For Barro and Lee (1994) all government spending must be financed by distortionary taxes that tend to undermine growth, so that the net effect of the government budget on growth depends on the supply-side contribution of the spending. Barro and Lee assume that defense and education spending have direct positive effects on productivity or the security of property rights and therefore impact on growth differently from other categories of expenditure. Thus the Barro-Lee ratio is expected to be negatively correlated with growth. An index based on the number of coups per year, the average number of revolutions, assassinations, and strikes per year, capture political instability. The index is expected to affect growth negatively.
Generally speaking, the evolution of the growth determinants usedin the two regression modelswas unfavourable for Sierra Leone,by comparison even with Sub-Saharan Africa (see Table 3). Between 1960 and 2000, life expectancy in Sierra Leone, reflecting health conditions, increased only modestly from 32 to 35 years, compared with 40 to 48 years for Sub-Saharan Africa. Average years of schooling attained by the population aged 15 or more in Sierra Leone was 0.66 in 1960 and 1.34 in 1960-90, compared with 1.13 and 1.53 for Sub-Saharan Africa. Sierra Leone’s real GDP per capita (1985 international prices) was US$876 in 1960 and US$905 in 1985 compared with US$733 and US$963 for Sub-Saharan Africa. On the whole real GDP per capita over 1960-97 was higher in Sierra Leone: US$ 996, than in Sub-Saharan Africa: US$876. An already high age dependency ratio of 77% in 1960-64 for Sierra Leone increased to 93% in 1990-97. Such a high age dependency ratio could lead to low financial savings, and undermine human capital accumulation by spreading out educational resources more thinly. Sierra Leone’s relatively low population growth rate of 2% compared with Sub-Saharan Africa’s 2.8% in 1960-97 suggests that population pressure might not account for much of Sierra Leone’s poor growth performance. Investment measured in constant international prices was extremely low in Sierra Leone, 1.3% of GDP for 1960-97, compared with 7.8% for Sub-Saharan Africa. A black market spread of 116% in the 1980s, and an inflation rate of 63% suggest an inauspicious policy environment for growth in the 1980s. Finally,the Barro-Lee unproductive government spending ratio was very high for Sierra Leone, 28% of GDP in 1960-84 compared with 15% for Sub-Saharan Africa (for 1960-97).
Tables 4a and 5a present the parameter estimates of the two models, and Tables 4b and 5b the corresponding predictions. Both models under-predict Sierra Leone’s pre-1990 growth, implying that for the observed values of the regressors, per capita GDP growth should have been worse than observed. The O’Connell and Ndulu model, with smaller residuals, gives better predictions. Its under-prediction is less than two standard errors of the residuals, ranging from 3.96 percentage points in 1980-84, to 0.85 percentage points in 1975-79. The Hoeffler model’s under-prediction ranges from 10 percentage points in 1960-64 to 4.4 percentage points in 1980-84, well over two standard errors of the residuals. Neither model conditions on civil war, and predictions for this period are in any case unavailable from the O’Connell-Ndulu model due to missing values for the included variables. Unsurprisingly, however, the Hoeffler model over-predicts growth during this period (1990-97) by 1.3 percentage points.[4]